Definitions

Electrical properties can be determined at various frequencies. The interaction between an electromagnetic wave and condensed matter can be described using permittivity and conductivity concepts. Fig. 25.1 shows the permittivity, a, versus frequency, v (or more usually ω= 2πν). In the high frequency region, there are various resonances arising from ionic (molecular) and electronic motions. The latter are usually described in terms of optical spectroscopy. In the low frequency region, on the other hand, dipolar and space charge relaxations, usually called dielectric relaxation, are expected. The space charge region, often observed in usual a.c. conductivity measurements, may be described in terms of a (complex) impedance formalism (see Chapters 26 & 27). These relaxations are directly related to the bulk conductivity and to electrode/electrolyte interfacial phenomena. They depend strongly on the microstructure on a 0.1–100 urn scale (porosity, surface topology) and on chemical reactions at the interface (polarization, diffusion). This low frequency domain corresponds to ‘free charges’, which can move in association with an alternative electric field, while the vibrational region at higher frequencies corresponds to ‘atom bonded fixed charges’ (‘dipoles’).

How are ions able to move in a solid? The standard answer to this question states that two different kinds of ionic motions can be discerned, namely oscillatory motion and jump diffusion (see Chapter 30). In fact, the motion is not only limited to oscillations and to statistical hopping from site to site.

Introduction

Proton conduction in β- and (β″-Al2O3 is of interest because of potential use in steam-electrolysis/H2/O2 fuel-cells. Single crystal H30+/-β″-Al2O3 was reported by Farrington & Briant to be a fast ion conductor with a proton conductivity of 10–4–10–5 (Ω cm)–1 at 25 °C. These encouraging results led to the first fabrication, characterization and use of polycrystalline H3O + -β″ /β-A12O32.

A second electrolyte with a higher conductivity but a lower thermal stability is NH4+ -β″-Al2032. This system, NH4+/H30+ -β″-Al203, is the best conductor of protons (σ = 7 × 10–4 (Ω cm)–1 rising to 2 × 10–2 (Ω cm)–1 at 200 °C). All previous work has been on single crystals (see Chapters 13 & 23) or powders. This chapter describes the synthesis of NH4+O + -β″-Al2O3 polycrystals.

Ion conducting structure of β″- and β-aluminas

The β-A12O3 family of compounds is constructed from oxygen and aluminium ‘spinel blocks’ with intervening ‘conduction planes’. In βA12O3 there are two spinel blocks per unit cell and in β″-A12O3 there are three. The upper oxygen layer of the spinel block is mirrored across the conduction plane in β-Al2O3 but this symmetry is lost in β″-A12O3. The alkali ion ‘Beevers-Ross’ site in the β-Al2O3 conduction planes is octahedral. The equivalent site in β″-A12O3 is tetrahedral and smaller. Thus K+ (r = 0.28 nm) promotes formation of β-A12O3 and Na+ (r = 0.196 nm), β″- A12O3.

General introduction

For solid protonic conductors the broad term ‘protonic species’ includes all those hydrogen-containing entities formed, perhaps only transiently during the conduction process, by ionic or molecular groups of sufficient basicity. This description includes the wide range of anionic protonated species in anhydrous protonic conductors: OH–, protonated tetrahedral ions such as HSeO4–, HPO42– and dimeric XO4 groups such as H(SeO4)23–, in addition to the cationic proton hydrates [H(H2O)n+, including H3O+, H5O2+, H7O3+ etc., and combinations of hydrogen with nitrogen: ammonium NH4+, hydrazinium N2H5+ and hydrazinium (2 +) (dihydrazinium or hydrazonium), N2H62+. Structural and spectroscopic data on the anionic protonic species encountered in anhydrous materials are included in later chapters and will not be referred to further here, except in cases where the counterion is ammonium or a proton hydrate.

Detailed information on the proton hydrate series in the solid state is available, and the reader is referred to articles reviewing structural studies and vibrational spectroscopic studies of proton hydrates, but also general overviews, the most recent being the comprehensive review of 1986.

The extensiveness of the proton hydrate series is not mirrored in the analogous nitrogen containing family. Of the isoelectronic pairs OH3+/NH4+ and O2H5+/N2H7+, in the solid state only the former is known.

Introduction

Some mixed conductors, i.e. electronic and ionic, into which ions can be rapidly and reversibly inserted can undergo a colour change. This is, for example, the case of the hydrogen ’bronzes‘, mentioned earlier in this book. The colour change can be either from transparent to coloured or from one colour to another. This phenomenon, which can be produced electrochemically, is called electrochromism. It is broadly defined as the production of an absorption band in a display material caused by an applied electric field or current. Such a property is currently under intensive study because of its potential use for passive information display, glare-free rearview mirrors for automotives, solar control windows or ‘smart’ windows, thermal sensors and projection systems if matrixable.

The electrochromic devices (ECDs) exhibit many attractive features and offer some superior display qualities compared with liquid crystal displays (LCDs) and light emitting diodes (LEDs). The use of ECDs might improve some of the disadvantageous properties of LCDs and LEDs. They typically show a good aesthetic appearance and a good colour contrast especially under high levels of ambient light where emissive displays lose contrast. They have a wide angle of view which is an advantage over most liquid crystal displays. They may exhibit a continuously variable intensity of coloration and possess memory in either the bleached or coloured state without power consumption. They need low driving voltage to operate. Moreover, large area devices can be fabricated as well as all solid-state displays.

Proton conduction in solids via hydrogen bonding was suggested in 1950 by Ubbelohde & Rogers. Later, the Grotthuss mechanism or translocation was proposed to explain the conductivity of para-electric potassium dihydrogen phosphate, KDP, and claimed soon after also for H3O + ClO4–, oxonium perchlorate, OP. Reviews that classified protonic superionic conductors, PSC, appeared from 1980 and were based successively on the ion-exchange properties of materials, their structures and their conduction mechanisms.

In this chapter, the first part (Sections 1.1–1.3) deals with the conduction mechanisms, while the second part (Sections 1.4 and 1.5) points out the significant structural and chemical factors leading to the different conduction mechanisms. The hydrogen bond is a common feature and serves as Ariadne's thread.

From ionic to protonic conduction

Ionic conduction

Ionic conductors may be divided into three classes depending on their defect concentrations (i) dilute point defects, dpd (∼ 1018 defects cm–3), (ii) concentrated point defects, cpd, (∼ 1O18–20 cm–3) and (iii) liquid-like or molten salt sublattice materials, mss, (∼ 1022 cm–3).

For dilute and concentrated point defect materials, examples are, respectively, NH4C;O4 (σ250°C = 10–9 Ω–1 cm–1, E = 1.4 eV) and CeF3 (σ24o°c = 6 × 10–4 Ω–1 cm–1, E = 0.26 eV). The conductivity occurs by an ion hopping mechanism.

In molten salt sublattice materials, practically all the ions in the sublattice are available for motion with an excess of available sites per cation as in e.g. Na β-alumina, (σ25°c = 1.4 × 10–2 Ω–1 cm–1, E = 0.16 eV).

Introduction

Some proton conductors have relatively high conductivities at room temperature. Introduction of these materials into electrochemical cells brings about attractive chemical sensors workable at room temperature. Potentiometric or amperometric detection of chemical components at room temperature would create new fields of application for sensors especially in bioprocess control and medical diagnosis. With an all-solidstate structure, the sensors would be compatible with micro-fabrication and mass production, and small power consumption associated with their ambient-temperature operation would be intrinsically suited for cordless or portable sensors.

As listed in Table 36.1, a good deal of research has been carried out so far on proton conductor-based gas sensors workable at ambient temperature. Various inorganic and organic ion exchangers, such as hydrogen uranyl phosphate (HUP), zirconium phosphate (ZrP), antimonic acid (AA), and NAFION membrane, have been utilized in the form of a disc, thick- or thin-film. The ionic conductivities of these proton conductors, in the range 10–4 – 10–2 S cm–1, are modest but seem to be still sufficient for chemical sensing devices.

A problem with these conductors is that humidity in the ambient atmosphere affects the conductivities and hence sensing characteristics when an amperometric mode of operation is adopted, as described later. The sensing electrodes have been provided mostly by Pt, and the counter electrodes by Ag, PdHx, TiHx, H0.35MoO3, Hx WO3, etc.

Introduction

Most transition-metal oxides are unreactive towards molecular hydrogen below elevated temperatures. However, dissociated hydrogen reacts topotactically with a wide range of binary and ternary transition-metal and uranium oxides at ambient temperature to produce hydrogen insertion compounds of formula HxMOn or HxMM′On which are often referred to as ‘hydrogen bronzes'. These compounds are, in general, non-stoichiometric, with biphasic regions containing solids of fixed composition in equilibrium, separated by single-phase regions of variable hydrogen content. Powder X-ray diffraction confirms that the lattice parameters of the parent oxide are little changed on hydrogen insertion, with the implication that the metal-oxygen framework is largely retained. (At high x values, however, amorphous products are sometimes found implying that a structural collapse has occurred.) The maximum hydrogen contents achieved under standard conditions are controlled by both structural factors and the redox characteristics of the metal oxidation states involved (Table 7.1).

The hydrogen insertion compounds HxMOn are formally mixed valence and, in marked contrast to the parent oxides, often have deep colours and behave as good electronic conductors, either metallic or semiconducting. The controllable variation in electronic properties has been exploited in electrochromic displays, which utilize the colour changes induced by insertion into e.g. WO3 or IrO2 films and in other sensors, which respond to conductivity or optical changes in oxides on hydrogen incorporation. The recovery of hydrogen on heating, HxMOn = Hx–yMOn + y/2H2 (for favourable systems at high x values), suggests a possible application as hydrogen storage materials.

Protonic conduction is encountered in a series of ‘composite’ compounds ranging from clays to ‘porous’ polymer structures via gels. In all these materials the distinction between the host lattice and protonic species becomes difficult because of the ‘softening’ of the rigid lattice. The rigid framework can be composed of big ions such as Keggin salts (d ∼ 2.5 nm), amorphous inorganic polymers (d ∼ 2.5–10 nm) or very small crystallites or particles such as the well-known clay minerals (d – 50–100 nm) (Fig. 18.1). The latter can also be prepared in the gel state if interaction with water is sufficiently strong. Some other materials, zirconium phosphate hydrate, Zr(HPO4)2.nH2O, for instance, can be obtained in the crystalline, amorphous and gel states. The formation of a gel depends on the existence of defects involving both the framework and associated protonic species and can be described in the following way: ‘dissolution’ of the host lattice leads to the formation of an associated water layer between solid and liquid, the increase of the solid–liquid interface being correlated. Here can also be cited the alteration phenomena of well-known silicoaluminates (mica, feldspar) in clays by proton exchange of first and second group cations. Fig. 18.2 summarizes the relationships existing between these materials and those discussed in the preceding chapters, ((β-alumina, NASICON and HUP). The differentiation between the structure and properties of ‘surface’ and ‘bulk’ remains valid if a chemically and thermally stable framework, i.e. of low ‘water’ content, is considered.

Introduction

Uranyl phosphates and arsenates appear interesting as uranium ores. There are about 200 known uranyl phosphate containing minerals which can be divided into three groups according to the UO2/PO4 ratio. The first group with UO2/PO4 = 1 comprises autunites and uranium phosphate micas; the second has UO2/PO4 > 1, most frequently 3/2 and 4/3, and is illustrated by phospharanglite; the third with UO2/PO4 < 1 is represented by parsonite. The first group of minerals contains usually Cu, Mn, Fe and H while in all groups Pb, Ca and Al are present. The formulae of these minerals have not been established accurately and errors in the determination of the UO2/PO4 ratio often lead to incorrect descriptions. Among the above minerals autunite, CaUO2PO4. 3–6H2O, appears to be the most extensively studied.

As far as hydrogen uranyl phosphate (HUP) is concerned, Beintema noticed – as early as 1938–that ‘true vagabond ions are present in the water layer of the structure’. Thirty years later, Sugitami et al. concluded on the grounds of NMR measurements that water molecules in KUO2PO4.3H2O are highly mobile. Much the same conclusions were reached by Wilkins et al. who observed a high mobility of oxonium ions in H3OUO2AsO4.3H2O. Thus oxonium and water protons diffuse rapidly in the structure of uranyl phosphate or arsenate. In spite of the experimental evidence the question of the true and bulk conductivity has been discussed for a long time.

During the past ten years, there have been numerous efforts to give an adequate description of superionic conduction, but they concern exclusively non-protonic conductors. In these compounds, neither tunnelling mechanisms nor oriented hydrogen bonds have been considered. However, in certain anhydrous compounds containing oxonium and ammonium ions, for instance, protonic conduction can also be interpreted in terms of such ‘non-protonic’ models. We shall start by giving a review of general models before discussing the problems particular to protonic conduction.

Theoretical interpretations of superionic conduction

Three main types of theoretical description can be distinguished, (i) Models derived from liquid or highly disordered solid electrolytes such as glasses or polymers. (ii) Continuous models based on solutions of Langevin's equation. These models can be improved by taking into account cooperative effects corresponding to liquid models (Brownian or stochastic particles) fitted to a periodic medium, (iii) Hopping models or lattice gas models, initially developed for phase transitions. They can consider dimensions higher than one and are closely associated with studies on reorientational motion in plastic crystals and on surface melting such as occurs with rare gas layers on graphite.

Fig. 30.1 shows that for the diffusion of a particle, several cases can be distinguished depending on the relative magnitudes of residence time τ0 on site A and of time of flight τ1 to site B. Existence of (polyatomic) protonic species establishing an oriented bond implies a necessary correlation between translational and rotational motions during a diffusive jump.

Introduction

Protonic conduction can be considered as a particular case of ionic conduction; however, there are some similarities with electronic conduction because of the proton size. In both cases, defects can play an important role.

Ionic conductivity in a compact structure is generally very low and requires a high activation energy. The diffusion coefficient of oxygen in amorphous silica at 1000°C, for instance, is 10–14cm2 s–1 and that in MgO is 10–20 cm2 s–1, with activation energies between 1 and 4 eV. In single crystal and polycrystalline alumina at 1700 °C, the diffusion coefficient decreases to 10–15 and 10–12cm2 s–1, respectively, and the activation energy increases to 6–8 eV. The diffusion coefficients of smaller cations such as Al3+ are of the order of 10–9–10–10 cm2 s–1 with activation energy of about 5 eV.

The diffusion coefficient of Al3+ ions in interstitial sites can be obtained from electrical conductivity measurements. Its activation energy is considerably lower (2.5 eV) than that of lattice diffusion. This applies not only to Al3+ in alumina but generally to all atoms occupying defect sites. All compounds have naturally a certain number of intrinsic defects as well as those associated with impurities (extrinsic defects). Intrinsic defects can be thermally activated and are particularly important for non-stoichiometric compounds. In the latter, diffusion coefficients are generally higher and the activation energy lower than that in stoichiometric compounds.

Triphosphate phases

Tetravalent metals readily form triphosphates of the type M(I)M(IV)2(PO4)3, where M(I) may be any alkali metal and M(IV) is most commonly Zr. A large number of such compounds have been synthesized by high temperature solid state methods. These compounds are of interest because they are low expansion materials and form the basis for an extensive family of solid electrolytes. Single crystals of NaZr2(PO4)3 have been grown from melts and the structure determined. A schematic drawing of a portion of the structure is shown in Fig. 15.1. The crystals are rhombohedral a = 8.8043(2) Å, c = 22.7585(9) Å, space group. The structure consists of zirconium octahedra and phosphate tetrahedra which are linked by corners. A basic unit consists of two octahedra and three tetrahedra separated by Na+ ions which form columns parallel to the c axis. Thus, the structure consists of [Zr2(PO4)3]– anions and Na+ cations. The anions are linked together through bridging phosphate groups to form a rigid framework which creates cavities. The cavity labelled A in Fig. 15.1 contains the Na+ which is coordinated by phosphate oxygens in a trigonal anti-prism. Only half the structure in the a-direction is shown in the figure. On adding the second half, three more cavities are created. Six such cavities, three above and three below, termed type II, surround the A type (type I) and are connected through narrow passageways.

Introduction

Vibrational (infrared, Raman and neutron) spectroscopy can give useful information about proton conductors, i.e. about structural as well as dynamical aspects. Infrared spectroscopy appears particularly suited for such studies since AH stretching modes (A = O, N, halogen) give rise to strong absorption bands in a region (4000-1700 cm–1) where there is not much interference from other groups. Various protonic species can thus be studied, even at very low concentrations.

As far as crystalline structures are concerned, X-ray and/or neutron diffraction methods and vibrational spectroscopy are complementary. The former can determine with accuracy the structure of the rigid framework but have some difficulties in locating protonic species, particularly if they are (statically or dynamically) disordered and if their concentration is low. The latter are well-suited for the identification of protonic entities and their types of association, the investigation of some structural details such as different crystallographic sites, (non)equivalent molecules, protonation sites, A–H and A … B distances and the nature and degree of structural disorder.

Spectroscopic data can also be used to study dynamical aspects, either as proton dynamics deciphered from the AH stretching band profile or as the dynamics of phase transitions. They are helpful in determining the order (first or second), the nature (displacive, order-disorder, reconstructive) and particularly the mechanism of the transformation at the molecular level. This can also shed some light on the conductivity mechanism, which can change considerably in going from one phase to the other.

Introduction

To a large extent the electrical properties of ice are similar to those of semiconductors: exponential increase in conductivity with increasing temperature, analogy of positive and negative carriers with electrons and holes, change of conduction upon doping. This similarity makes it possible to consider, in the case of ice, models for numerous devices that have been fabricated using semiconductors. However, it can hardly be expected that ice-based devices will have any advantage (qualitative, technological) against semiconductor analogues. Therefore, consideration of such icebased devices is of no practical significance.

On the other hand, the electrical properties of ice, principally due to the presence of a large number of carriers and the nature of the carriers, may give rise to memory effects which are much more complicated compared with the electrical properties of ordinary semiconductors. Therefore, using ice, one can realize devices which either have no analogues amongst semiconductor devices or which are advantageous compared with the latter. Some such devices are described next.

Before giving detailed descriptions, we shall emphasize two important features. Firstly, we are dealing here with models rather than with operational devices. Secondly, in selecting concrete models, we are basing our consideration on the most interesting physical properties of ice, treated in original research papers.